See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/314843576

The Namib-Thar Desert Disjunction in Dactyliandra () is the Result of a Recent Introduction to India

Article in Systematic · March 2017 DOI: 10.1600/036364417X694548

CITATION READS 1 281

4 authors, including:

Arun Pandey Hanno Schaefer University of Delhi Technische Universität München

74 PUBLICATIONS 271 CITATIONS 138 PUBLICATIONS 2,708 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Plants & Pollinators View project

Systematics of Alysicarpus, Crotalaria and Indigofera (Fabaceae) View project

All content following this page was uploaded by Hanno Schaefer on 13 March 2017.

The user has requested enhancement of the downloaded file. Systematic Botany (2017), 42(1): pp. 63–72 © Copyright 2017 by the American Society of Taxonomists DOI 10.1600/036364417X694548 Date of publication March 1, 2017 The Namib-Thar Desert Disjunction in Dactyliandra (Cucurbitaceae) is the Result of a Recent Introduction to India

Katharina Lindner,1 Yue Hu,1 Arun K. Pandey,2 and Hanno Schaefer1,3 1Plant Biodiversity Research, Department Ecology and Ecosystem Management, Technical University of Munich (TUM), Emil-Ramann Str. 2, D-86354 Freising, Germany. 2Department of Botany, University of Delhi, Delhi-110 007, India. 3Author for correspondence ([email protected])

Communicating Editor: Chrissen E. C. Gemmill

Abstract—Depending on taxonomic concept, the genus Dactyliandra (Cucurbitaceae) comprises up to four species: two from Southwest Africa, D. luederitziana (Namibia) and D. welwitschii (), and two very locally distributed and poorly known taxa from Northeast Africa, D. nigrescens and D. stefaninii. In addition to these African populations, D. welwitschii was also reported as a possibly native species from the Thar desert of Northwestern India and Pakistan ca. 8,000 km away from the Namib range. A phylogenetic analysis of plastid and nuclear ribosomal ITS DNA sequences including all Dactyliandra species and D. welwitschii from the Thar desert revealed that this striking disjunction is of very recent origin. Dactyliandra is monophyletic only after exclusion of the East African taxa, which are best placed in . The species Dactyliandra welwitschii is monophyletic and includes the Indian , which group with a sample from an Angolan population. We hypothesize that this unexpected connection is a result of man-mediated introduction to India from Africa perhaps in the times of the slave trade. The new combinations Blastania lucorum and Trochomeria nigrescens are published here. Keywords—Blastania, continental tracks, molecular clock, Saharo-Sindian desert belt, Siddi, Trochomeria.

The genus Dactyliandra Hook.f. was originally described by Africa. A further East African species, D. stefaninii (Chiov.) the British botanist Joseph D. Hooker in 1871 from material C. Jeffrey, also remained more or less unnoticed. Dactyliandra collected in Angola, southwestern Africa (Hooker 1871). The stefaninii had been described by Chiovenda as a species of species was described as D. welwitschii Hook.f. to honor the Wight & Arn. (Chiovenda 1917), it was later trans- Austrian explorer and botanist Friedrich Welwitsch (1806– ferred to Trochomeria Hook.f. by Jeffrey in 1962 and finally 1872) who had collected the type material. Further fieldwork moved to Dactyliandra in 1985 (Jeffrey 1962, 1985). These East revealed that its distribution is restricted to the Namib region African plants reduced the extreme disjunction in the genus of Angola and adjacent Namibia. The Namibian plants (Fig. 1) but at the same time called into question the species circum- seemed to be morphologically slightly different (3-lobed leaves scriptions in Dactyliandra, with the populations of the two instead of 5- to 7-lobed leaves) when Cogniaux studied the extreme ends of the range being attributed to the same spe- very limited number of herbarium collections available to him cies and two different species living more or less in sympatry and therefore described them as a second species, D. luederitziana in between (Fig. 2). Cogn. (Cogniaux 1916). This concept, however, was not Family-wide phylogenetic analyses placed D. welwitschii accepted and the following authors treated all Southwest in the tribe in a well-supported clade together African plants as one single species D. welwitschii (e.g. Meeuse with representatives of the genera Trochomeria, Blastania 1962, Gibbs Russell et al. 1987). Kotschy & Peyr. (syn. Hook.f.), and Benincasa Savi It was a big surprise when almost a century after Welwitsch’s (Schaefer et al. 2009; Schaefer and Renner 2011). However, discovery, the Indian botanist M. Bhandari found Dactyliandra these analyses only included one representative per genus, plants morphologically matching D. welwitschii in Rajasthan so could not elucidate the evolution and historical biogeog- about 8,000 km away from the species’spreviouslyknown raphy within Dactyliandra. distribution range (Bhandari and Singh 1964; Fig. 2). In Here, we use a comprehensive sample of Dactyliandra and Rajasthan, the species was reported to be locally common as a presumed relatives to test in a phylogenetic framework climber over Capparis L. bushes and in hedges but Bhandari & the hypotheses that 1) Dactyliandra is monophyletic, and Singh (1964) also suggested that it might have been 2) populations in India/Pakistan, East Africa, and Southwest overlooked or confused with the similar looking Blastania Africa are three genetically and morphologically distinct enti- garcini (L.) Cogn. (syn. Ctenolepis garcini (L.) C. B. Clarke) and ties and remainders of a former widespread ancestor, whose might be more widespread. Later, the species was indeed also range covered the entire Saharo-Sindian desert belt region. found in the Kirthar Range on the Pakistani side of the border (Khatoon 2006). This suggested that the total Thar region pop- Materials and Methods ulation might be bigger and cover a relatively large range, which could indicate an indigenous status. The case was Taxon Sampling — For the molecular phylogenetic analyses, we included widely cited as an example of an extreme disjunction and even seven accessions of Trochomeria and 15 accessions representing the three or used by panbiogeographers as evidence for “tracks” between four Dactyliandra species: ten specimens of D. welwitschii/D. luederitziana, four of them from Namibia, and three from India and Angola respectively, continents (Craw et al. 1999). However, doubts about its native and we downloaded sequences for an additional Namibian sample from status remained and a human-mediated introduction to the GenBank. For D. stefaninii and D. nigrescens, we sequenced two specimens Thar region could not be excluded. each, all of East African origin. For the presumed sister genus Trochomeria, Less attention was given to an additional species of we included the following four species in our phylogenetic analyses: T. debilis Benth. & Hook.f. (one specimen), T. macrocarpa Harv. (three speci- Dactyliandra, D. nigrescens C. Jeffrey (Jeffrey 1965), which was mens), T. polymorpha Cogn. (two specimens), and T. sagittata Benth. & Hook.f. described at the same time from Eastern Africa, roughly in (one specimen). Finally, we added five samples of the genera Blastania and the middle of the route from Northern India to Southwestern Benincasa as outgroups based on the results of earlier family-wide analyses

63 64 SYSTEMATIC BOTANY [Volume 42

Fig. 1. Dactyliandra welwitschii Namibia, Farm Kyffhauser. A. Habit; note simple tendrils and 5-lobed leaf. B. Male flower and ciliate bracts; note glandular hairy petal surface. C. Unripe green fruits. D. Mature pink fruits. Scale bars = 5 mm. (Photographer A. Dreyer).

(Schaefer et al. 2009, Schaefer and Renner 2011). Appendix 1 provides (Stocks) A. Meeuse, 26 of Trochomeria macrocarpa (Sond.) Hook.f., five of voucher information and GenBank numbers for all included sequences. T. polymorpha Cogn., and one of T. debilis Hook.f. (for specimen details For the morphological comparisons and development of the determina- see appendix). Further specimens from the herbarium of the Botanische tion key, we studied 115 specimens at the herbarium of Royal Botanical Staatssammlung Munich (M) were studied to see if morphological Gardens Kew (K): for the genus Dactylindra, we studied 52 specimens of variation was within the range of variation of the vouchers in K. All D. welwitschii/D. luederitziana, eight specimens of D. stefaninii, and eight measurements were taken from dry herbarium specimens. specimens of D. nigrescens;15specimensofBlastania cerasiformis Scans of type material of all Dactyliandra species can be accessed in the Global plants database of JSTOR using the following links: http://plants.jstor.org/compilation/Dactyliandra.welwitschii http://plants.jstor.org/compilation/Dactyliandra.nigrescens http://plants.jstor.org/compilation/Dactyliandra.stefaninii DNA Extraction and PCR — DNA extraction followed standard pro- cedures described in Kocyan et al. (2007). Polymerase chain reactions (PCR) were performed using two different premixed reaction kits: a) the Nova Taq™ PCR Master Mix (Merck) and b) the KAPA2G™ Fast 2x ReadyMix (Kapa Biosystems Inc.). We amplified the nuclear ribosomal internal transcribed spacer (ITS) region using the primers of Douzery et al. (1999), and from the plastid genome the rbcL and matK genes, the rpl20-rps12 spacer, the trnL intron and trnL-trnF spacer with the primers listed in Kocyan et al. (2007), and the trnH-psbA spacer with the primers of Sang et al. (1997). The PCR products were cleaned with the ExoSAP- IT® enzyme mix. Then we sent the cleaned PCR products premixed with sequencing primer to a private company (GATC Biotech AG, Konstanz, Germany) for Sanger sequencing. Each PCR product was sequenced in two directions and the sequences edited and assembled using Geneious v. 6.0 (Biomatters Ltd., Auckland, New Zealand). Eigthy-nine sequences were generated for this study. An additional 17 sequences were downloaded from GenBank (www.ncbi.nlm.nih.gov/genbank/), most from Fig. 2. Distribution map showing the natural ranges of the species of our previous family-wide analyses (Kocyan et al. 2007, Schaefer et al. 2009). Dactyliandra and Trochomeria based on all label information of all herbar- Appendix 1 lists the relevant taxonomic names with authors, vouchers, geo- ium material seen. graphic origin, and GenBank accession numbers (KU289638–KU289736). 2017] LINDNER ET AL.: BIOGEOGRAPHY OF DACTYLIANDRA 65

Sequence Alignment and Phylogenetic Analyses — Sequences were three from Angola, and three from India. Furthermore, there aligned using the Geneious alignment algorithm implemented in Geneious. is one sequence of D. stefaninii from Ethiopia and three Unambiguous gaps in the ITS, trnH-psbA,andrpl20-rps12 alignments were coded manually as binary characters following Simmons & Ochoterena Trochomeria plus three Blastania sequences. Unfortunately, all (2000) and added as additional columns to the three alignments. Maxi- attempts to amplify and sequence ITS for the remaining mum likelihood (ML) analyses (and non-parametric bootstrap searches samples failed. The D. stefaninii sequence is very similar to with the fast-bootstrap algorithm) were performed using RAxML-VI-HPC the sequence of Trochomeria macrocarpa from Burkina Faso (Stamatakis et al. 2008) on the CIPRES science gateway (Miller et al. 2010). (two nucleotides different) but differs considerably from all Based on the Akaike Information Criterion (Akaike, 1974) as implemented in jModeltest (Posada 2008) we selected the GTR + G model (six general D. welwitschii sequences. The Namibian D. welwitschii sam- time-reversible substitution rates, assuming gamma rate heterogeneity), ples are represented by two ribotypes differing in three posi- with model parameters estimated over the duration of specified runs. tions. Among the three Angolan samples, we also found two Analyses in RAxML were first run on all six alignments (including the ribotypes differing in nine positions. All three Indian sam- coded gap columns) independently. We then tested congruence of the five plastid regions and the ITS data using the Partition Homogeneity test in ples have identical ITS sequences, which are very similar to PAUP* version 4.0a149 (Swofford 2002) and by manual comparison of the one of the Angolan ribotypes (Carrisso and Sousa 3/8) differ- individual topologies. No incongruence was detected, so all sequenced ing only in the number of repeats in a poly guanine zone regions were concatenated into a single final alignment. at position 184 of our alignment (six repeats instead of five). — Divergence Time Estimation Divergence times were estimated using Sequence Analyses of the Plastid Regions—Among the the Bayesian relaxed clock approach implemented in BEAST version 1.8.2 (Drummond and Rambaut 2007). Following the BEAST manual (Drummond five chloroplast DNA regions amplification success was highest and Rambaut 2007), we modified our final combined sequence dataset for the trnH-psbA spacer region, where our alignment includes slightly: we merged identical sequences and deleted the coded indel columns 18 sequences of a total aligned length of 266 nucleotides plus from the matrix. Three clades were defined a priori based on the results of four coded gaps. The four East African Dactyliandra sequences our ML analyses and previously published phylogenies (Kocyan et al. 2007; Schaefer et al. 2009; Schaefer and Renner 2011): (a) Benincasa,(b)Blastania, (one from D. nigrescens, three from D. stefaninii)areallquite (c) Blastania + Trochomeria + Dactyliandra.Intheabsenceoffossilsforour different from D. welwitschii (up to six differences) and appear clade of interest, we followed a secondary calibration approach and most similar to Trochomeria debilis and T. sagittata. For constrained divergence times for three nodes and the root of our tree using D. welwitschii, we obtained four sequences from Namibian the results of a previous family-wide analysis (Schaefer et al. 2009). Specifi- samples, one from Angola and three from the Indian material. cally, we set the split Benincasa hispida (Thunb.) Cogn. - B. fistulosa (Stocks) H.Schaef. & S.S. Renner (syn. Praecitrullus fistulosus (Stocks) Pangalo) to All Namibian sequences are identical and differ from the 5 ± 2 Ma (million years ago), the split Blastania cerasiformis - Bl. lucorum to Angolan sequence in two positions. The three Indian samples 7 ± 3 Ma, the split Blastania vs. Trochomeria plus Dactyliandra to 10 ± 3 Ma, represent two different haplotypes, differing from the Angolan and the root of our tree (Benincasa vs. all other taxa) to 13 ± 3 Ma, all with and Namibian sequences in one to three positions. a normal prior distribution. Searches used a Yule tree prior, the GTR + G (general time reversible plus gamma) substitution model, and 50 million For rbcL, our alignment includes 16 sequences with a total Markov chain Monte Carlo (MCMC) generations, sampling every length of 1,375 nucleotides. We were unable to produce an 1,000th generation. Tracer version 1.5 (Rambaut and Drummond 2009) rbcL sequence for any of the East African Dactyliandra sam- was used to check that effective sampling sizes had all reached > 200, ples, but for D. welwitschii, we obtained four sequences from suggesting convergence of the chains. TreeAnnotator, part of the BEAST Namibian material, two from Angolan samples and two for package, was then used to create a maximum clade credibility tree after removal of a burn-in of 10,000 trees from each of the four independent India (the latter only ca. 800 nucleotides long). These BEAST searches. The resulting chronogram was visualized using FigTree sequences are all identical except for one variable site (position v.1.4.0. (http://tree.bio.ed.ac.uk/software/figtree/). 844), where one Angolan and one Namibian accession share a — Biogeographic Analysis To estimate the ancestral range of the lineage cytosine, whereas the two Indian accessions share a thymine and reconstruct the origin of the Thar desert population of Dactyliandra, we coded species’ geographic ranges as an unordered multistate character, with the second Angolan sequence (Carrisso and Sousa 3/8). using the following character states: i) Asia, ii) Namibia, iii) Angola, Our matK alignment includes 15 sequences of a total iv) East Africa, and v) Madagascar. We then imported the highest likeli- aligned length of 1,143 nucleotides. Again, we had no amplifi- hood tree with branch lengths into Mesquite 3.04 (Maddison and cation success for the East African samples but managed to Maddison 2009) and used the Markov k-state one-parameter model get seven D. welwitschii sequences, three for Namibia and two (Lewis 2001), which assumes a single rate for all character state transitions to infer the probabilities of shifts between biogeographic regions. each for Angola and India. The Namibian sequences are iden- tical and differ from the Angolan and Indian sequences in three positions. Of the two Angolan sequences, one (Carrisso Results and Sousa 3/8) is identical to the two Indian sequences, the second differs in one position. Morphology—The two East African species Dactyliandra For the trnL intron and trnL-trnF spacer region our align- nigrescens and D. stefaninii differ from the remaining species ment contains 19 sequences with a total aligned length of in the genus in their sexual system (both dioecious) and in anther 957 nucleotides. One D. stefaninii sequence from Ethiopia was morphology with S-shaped thecae instead of hippocrepiform obtained and appears most similar to Trochomeria macrocarpa thecae. None of the studied leaf or flower characters allowed sequences but very different from D. welwitschii. For the to consistently separate Indian D. welwitschii from Angolan latter, five sequences of Namibian material are identical and D. welwitschii or from Namibian material labelled as differ from the one Angolan and three Indian sequences in D. luederitziana. The leaf characters given by Cogniaux (1916) just one position (581), where the latter have a guanine instead were found to be inadequate to separate the two species of a thymine. described from Southwest Africa, so we treat D. luederitziana Finally, the rpl20-rps12 spacer alignment contains 15 sequences in the following as a synonym of D. welwitschii. with a total aligned length of 745 nucleotides plus five coded Sequence Analysis of the Nuclear ITS Region—The ITS gaps. The sequences for D. nigrescens and D. stefaninii alignment includes 19 sequences with a total length of are again more similar to Trochomeria sequences than to 817 aligned nucleotides plus 11 coded gaps. Dactyliandra is D. welwitschii. The five Namibian D. welwitschii sequences are represented by five D. welwitschii accessions from Namibia, identical. The one Angolan sequence is identical to the two 66 SYSTEMATIC BOTANY [Volume 42 from India and both differ from the Namibian sequences in supported as a clade and two East African Dactyliandra samples six positions. seem to group as a separate clade outside Trochomeria. Blastania Phylogeny Inferences—The ML tree for the ITS alignment is again sister to all Trochomeria and Dactyliandra samples. (Fig. 3) shows a highly supported D. welwitschii clade, with Finally, the analysis of the combined ITS and plastid align- two Angolan samples branching first and then a sister ments resulted in a phylogeny where Blastania is sister to a relationship of a highly supported Namibian clade with a D. welwitschii clade (Angola (Angola plus India, Namibia)) third highly supported clade of the Indian samples plus one plus Trochomeria with inclusion of D. stefaninii and D. nigrescens from Angola. The genus Trochomeria does not appear to (Fig. 5). However, the bootstrap support for the Trochomeria be monophyletic but rather a grade with T. polymorpha clade is low. branching off first, then followed by a clade with T. sagittata Chronogram—The divergence time analysis resulted in a and T. macrocarpa from Burkina Faso, which is sister to the chronogram with a well-supported topology identical to the only East African Dactyliandra for which we obtained an ITS combined ML tree (Fig. 6). The 95% HPD (highest posterior sequence (D. stefaninii from Ethiopia). Blastania is sister to all density) for the divergence time estimate for the split Trochomeria and Dactyliandra samples. between D. welwitschii and the Trochomeria clade is 10.8– For the combined plastid alignment, the ML phylogeny 6.1 Ma (best estimate 8.5 Ma), for the crown node of (Fig. 4) is similar in topology to the ITS tree. Dactyliandra is D. welwitschii it is 6.4–2.4 Ma (best estimate 4.3 Ma), and for monophyletic when the East African species are excluded. One the split between the Namibian and the Angolan/Indian Angolan sample groups with the Indian samples and is sister clade it is 4.6–1.5 Ma (best estimate 3 Ma). The age of the to a well-supported Namibian clade. Trochomeria is not well- Indian samples must be younger than the split between the

Fig. 3. Maximum likelihood (ML) tree for Dactyliandra and relatives based on 817 nucleotides of the nuclear rDNA ITS 1, the 5.8S gene, and ITS 2. ML bootstrap support ≥50% given at the nodes. Native range indicated by colour: red = India, green = Namibia, blue = Angola, orange/yellow = East Africa, black = other. 2017] LINDNER ET AL.: BIOGEOGRAPHY OF DACTYLIANDRA 67

Fig. 4. ML tree for five plastid DNA regions combined (4495 nucleotides). ML bootstrap support ≥50% given at the nodes. Native range indicated by colour: red = India, green = Namibia, blue = Angola, orange/yellow = East Africa, black = other.

Angolan specimen Carrisso & Sousa 3/8 (HS1796) and the three analysis, due to the poor amplification success with the old Indian samples estimated to 1.7–0.16 Ma. East African material and incomplete sampling of Trochomeria. Biogeography—The geographic origin of the Dactyliandra/ Fieldwork in East Africa for collection of ecological and mor- Trochomeria clade is estimated to be in East Africa (Fig. 7). The phological data and fresh leaf material for good quality DNA ancestral range of D. welwitschii is reconstructed as Angolan, is needed to obtain a more definite answer for this question. from where the plants colonized both Namibia and India. We find no support for a split of the Namib population in an Angolan D. welwitschii and a Namibian D. luederitziana as Discussion suggested by Cogniaux (1916). Biogeography—Our results reject the hypothesis of a Taxonomy—The morphological and molecular data both widespread ancestor and three long isolated refugial areas indicate that Dactyliandra welwitschii is monophyletic but for the genus Dactyliandra. Instead, the most likely explana- not the genus Dactyliandra as a whole. The two East African tion for the striking 8000 km disjunction between the Namib species that seemed to be a possible link between the region and Rajastan seems to be a recent introduction of Namib and Thar desert populations (Jeffrey 1965) are in fact D. welwitschii from Angola to Northern India or Pakistan. better placed in the genus Trochomeria as T. stefaninii (Chiov.) We acknowledge that our divergence time estimate seems a C. Jeffrey and T. nigrescens (C. Jeffrey) H. Schaef., a new bit too old to be explained by human-mediated transport. combination for the latter is formalized in our key below. However, molecular clock estimates, especially those based Our results support maintaining two different species in East on secondary calibrations, are not very precise. Moreover, Africa, even though their precise relationship and position our sampling of Dactyliandra from Angola is very small, within Trochomeria could not be definitely resolved in our so we have probably not found the closest relatives of the 68 SYSTEMATIC BOTANY [Volume 42

Fig. 5. ML tree for ITS plus five plastid DNA regions combined (5323 nucleotides). ML bootstrap support ≥50% given at the nodes. Native range indicated by colour: red = India, green = Namibia, blue = Angola, orange/yellow = East Africa, black = other.

Indian plants. Adding them would reduce the divergence living in parts of India and Pakistan, including the Gujarat time estimate even further. In any case, a recent divergence region, where Dactyliandra is found. Shah and colleagues of one million years or less does not fit to the idea of an conclude that the Siddi people have been brought to India ancient disjunction. as slaves mainly by the Portuguese from their former African The results of the ancestral range estimation clearly show colonies, which includes Angola. However, the earliest that an Angolan origin of the Dactyliandra population in records of Siddi people in India are much older and date to India/Pakistan is most likely (Fig. 7). When searching for 1100 AD (Shah et al. 2011 and references therein). We there- possible connections between Angola and Rajasthan, we fore suggest as one possible explanation for the disjunction found no evidence for bird dispersal. The small pink berry of D. welwitschii an introduction with African slaves from fruit (Fig. 1D) is probably eaten by birds but at least today, Angola to Gujarat. A similar human-mediated dispersal sce- there are no bird migration routes from Angola to Rajasthan. nario has been suggested to explain the tropical America – External transport like in the cucurbit genus Sicyos (Sebastian West Africa disjunction in sphaerocarpa (Cogn.) et al. 2012) seems impossible because of the lack of any struc- H. Schaef. & S.S. Renner (previously known as Cucumeropsis tures on the seed surface that would attach the seeds to legs mannii Naudin), which might have been introduced from or feathers of a bird. Regarding possible human transport, Brazil to West Africa during the times of the slave trade there seem to be no links between the two regions today. (Schaefer and Renner 2010). We searched for ethnobotanical However, Shah et al. (2011) analysed in a recent study the evidence for medicinal usage of D. welwitschii in Namibia origin of the Siddi people, a group of African origin today and Angola, and also in the Thar region. All we found was 2017] LINDNER ET AL.: BIOGEOGRAPHY OF DACTYLIANDRA 69

Fig. 6. Chronogram for Dactyliandra and relatives based on the combined nrITS and plastid sequences estimated using BEAST version 1.8.2 (Drummond and Rambaut 2007). Scale bar in million years before present, numbers at the nodes are best age estimates, blue bars indicate the 95% CI ranges. that the species is listed in a register of Indian medicinal plants (Tandon and Sharma 2009). However, no actual usage is indicated in this register; only the high content of linoleic acid in its seeds is mentioned. This questions the assumed medicinal importance, but the knowledge about its medicinal properties could have been lost or Dactyliandra could have been replaced by more powerful drugs. What we know about the species habitat in India would also fit to a history of active introduction followed by escape from cultivation: in a recent field survey, Joshi et al. (2013) found only 38 individuals in four places in “agricultural hedges covered by dried thorny twigs and branches” on sand. They did not find D. welwitschii in natural habitats and this is similar to earlier reports from hedges around cultivated areas (Bhandari 1978, p. 154). Taken together with the low genetic diversity in the Indian population compared to the African plants, all this indi- cates that D. welwitschii is not indigenous to the Thar region and the striking disjunction is just the result of human activities. We cannot provide hard evidence that the slave trade is responsible for introduction of Dactyliandra to India, but based on our results, the hypothesis cannot be Fig. 7. Ancestral range reconstruction for Dactyliandra and relatives rejected either. More fieldwork to gather additional plant based on the combined nrITS and plastid sequences estimated using Mesquite version 3.0.4 (Maddison and Maddison 2009). Regions of origin material and also direct ethnobotanical evidence from both are color-coded: India = red, Namibia/South Africa = green, Angola = Southwest Africa and the Thar region is needed to defi- blue, East Africa = orange, Madagascar = black, ambiguous = grey, nitely answer the question. not coded (outgroups) = white. 70 SYSTEMATIC BOTANY [Volume 42

Key to Species of DACTYLIANDRA, TROCHOMERIA, and BLASTANIA

This determination key is based on our own measurements plus literature data (Jeffrey 1965, 1967, 1985). It includes all species of Dactyliandra, all species of Blastania and all accepted species of Trochomeria except the poorly known T. baumiana Gilg, T. teixeirae R. Fern & A. Fern, and T. wyleyana Benth. & Hook.f. for which we could not find sufficient material. For the genus Blastania (usually referred to as Ctenolepis), we here follow Kartesz & Gandhi (1994), who concluded that Blastania (Stocks) Kotschy & Peyr. has priority because it was published in July 1867 (in Kotschy and Peyritsch’s Plantae Tinneanae, 1867) while Ctenolepis (Stocks) Hook.f. was published in September 1867 (Bentham and Hooker’s Genera Plantarum, 1867).

1. Plant monoecious, with male and female flowers on the same individual...... 2 2. Bract margin dentate to shortly ciliate, the cilia length not exceeding half the bract’s diameter...... Dactyliandra welwitschii Hook.f. 2. Bract margin long ciliate, the cilia at least as long as the bract diameter...... 3 3. Male and female inflorescences very short and concealed by bract (India and )...... Blastania garcinii (Burm.f.) Cogn. 3. Inflorescences longer, with pedicels of female flowers and peduncles of male racemes ≥ 7 mm, inflorescence not concealed by bract (subtropical and tropical Africa, India and Pakistan)...... Blastania cerasiformis (Stocks) A.Meeuse 1. Plant dioecious, with male and female flowers on separate plants...... 4 4. Species confined to Madagascar...... Blastania lucorum (Keraudren) H.Schaef. comb. nov. (basionym: Zombitsia lucorum Keraudren, Adansonia ser. 2, 3:167. 1963) 4. Species from the African mainland...... 5 5. Tendrils absent...... Trochomeria polymorpha Cogn. 5. Tendrils present...... 6 6. Stipuliform bract absent...... 7 7. Male receptacle tube ± cylindrical, ≥ 13mmlong...... Trochomeria macrocarpa (Sond.) Hook.f. 7. Male receptacle tube ± campanulate, ≤ 4mmlong...... 8 8. Leaf blade broadly ovate-reniform, palmately 3-lobed, central lobe largest, lobes obovate-rhombic...... Trochomeria stefaninii (Chiov.) C. Jeffrey) 8. Leaf blade variable in outline. Lobes always deeply sagittate and acute, leaf length and breadth and number of lobes variable...... Trochomeria sagittata Cogn. 6. Stipuliform bract present (at least in a few axils)...... 9 9. Male receptacle tube ± cylindrical, ≥ 10mmlong...... 10 10. Corolla lobes rotate, angled, not strongly recurved and not exceeding half the receptacle tube length (Southern Africa)...... Trochomeria debilis Hook.f. 10. Corolla lobes of mature flowers strongly recurved...... 11 11. Corolla lobes exceeding half the receptacle tube length (sub-Saharan Africa)...... Trochomeria macrocarpa (Sond.) Hook.f. 11. Corolla lobes not exceeding half the receptacle tube length (Mozambique, South Africa, Swaziland)...... Trochomeria hookeri Harv. 9. Male receptacle tube campanulate, ≤ 2.5mmlong...... 12 12. Rachis of raceme geniculate (zig-zag shaped)...... Trochomeria stefaninii (Chiov.) C. Jeffrey 12. Rachis not geniculate...... 13 13. Fruit globose, not beaked, testa distinctly pitted...... Trochomeria stefaninii (Chiov.) C. Jeffrey 13. Fruit shortly beaked and testa smooth...... Trochomeria nigrescens (C. Jeffrey) H. Schaef. comb. nov. (basionym: Dactyliandra nigrescens C. Jeffrey, Kew Bull. 19: 221. 1965)

Acknowledgments. We thank A. Dreyer, Farm Kyffhauser, Namibia Drummond, A. J. and A. Rambaut. 2007. BEAST: Bayesian evolutionary for allowing us to use his photographs of D. welwitschii, S. Schepella for analysis by sampling trees. BMC Evolutionary Biology 7: 214. lab work, the curators of the K and M herbaria for allowing us to study Gibbs Russell, G. E., W. G. M. Welman, E. Retief, K. L. Immelman, their collections, and M. Romeiras for photographs and samples of LISC G. Germishuizen. G., B. J. Pienaar, M. Van Wyk, and A. Nicholas. specimens. Arun Pandey thankfully acknowledges grant support from 1987. List of species of southern African plants. Memoirs of the SERB-DST, New Delhi (#SR/SO/PS/116/2010). Botanical Survey of South Africa 2(1–2): 1–152 (pt. 1), 1–270 (pt. 2). Pretoria: Botanical Research Institute. Hooker, J. D. 1871. Cucurbitaceae. Pp. 521–569 in Flora of Tropical Africa Literature Cited vol. 2, ed. D. Oliver. : L. Reeve. Jeffrey, C. 1962. Notes on Cucurbitaceae, including a proposed new Akaike, H. 1974. A new look at the statistical model identification. IEEE classification of the family. Kew Bulletin 15: 337–371. Transactions on Automatic Control 19: 716–723. Jeffrey,C.1965.FurthernotesonCucurbitaceae.Kew Bulletin 19: Bentham, G. and J. D. Hooker, 1867. Genera plantarum, vol. 1. London: 215–223. Reeve. Jeffrey, C. 1967. Cucurbitaceae. Pp. 1–157 in Flora of Tropical East Africa, Bhandari, M. M. 1978. Flora of the Indian Desert.Jodhpur,India: eds. E. Milne-Redhead and R. M. Polhill. London: Royal Botanic M. P. S. Repros. Gardens, Kew. Bhandari, M. M. and D. Singh. 1964. Dactyliandra (Hook.f.) Hook.f.: A Jeffrey, C. 1985. Further Notes on Cucurbitaceae: VII. Preliminary to the cucurbitaceous genus new to the Indian flora. Kew Bulletin 19: 133–138. Flora of Ethiopia. Kew Bulletin 40: 209–211. Chiovenda, E. 1917. Plantae novae vel minus notae e regione aethiopica. Joshi, P. N., H. B. Soni, S. F. Wesley Sunderraj, and J. Joshua. 2013. Distri- Bullettino della Società Botanica Italiana 1917: 53–61. bution and conservation of less known rare and threatened plant Cogniaux, A. 1916. Cucurbitaceae-Fevilleae et Melothrieae. Pp. 1–275 in Das species in Kachchh, Gujarat, India. Our Nature 11: 152–167. Pflanzenreich vol. IV, ed. A. Engler. Leipzig: W. Engelmann. Kartesz, J. T. and K. N. Gandhi. 1994. Nomenclatural notes for the North Craw, R. C., J. R. Grehan, and M. Heads. 1999. Panbiogeography – Tracking American flora XIII; Cucurbitaceae. Phytologia 76: 444. the history of life. Oxford: Oxford University Press. Khatoon, S. 2006. Dactyliandra (Hook.f.) Hook.f. (Cucurbitaceae) – anew Douzery, E. J. P., A. M. Pridgeon, P. Kores, H. P. Linder, H. Kurzweil, generic record from Pakistan. International Journal of Biology and and M. W. Chase. 1999. Molecular phylogenetics of Diseae Biotechnology 3: 819–820. (Orchidaceae): A contribution from nuclear ribosomal ITS Kotschy, T. and I. Peyritsch 1867. Plantae Tinneanae. Vienna, Typis Caroli sequences. American Journal of Botany 86: 887–899. Gerold filii. 2017] LINDNER ET AL.: BIOGEOGRAPHY OF DACTYLIANDRA 71

Kocyan, A., L. B. Zhang, H. Schaefer, and S. S. Renner. 2007. A multi- DQ535750, DQ536669, *, DQ536810, DQ536545, HQ201973, Dactyliandra locus chloroplast phylogeny for the Cucurbitaceae and its implica- welwitschii Hook. f., DNA no. HS709, unvouchered, Namibia, KU289703, tions for character evolution and classification. Molecular Phylogenetics KU289732, KU289717, KU289693, KU289656, KU289642, KU289681, and Evolution 44: 553–577. Dactyliandra welwitschii Hook. f., DNA no. HS892, Seydel 990 (US), Lewis, P. O. 2001. A likelihood approach to estimating phylogeny from Namibia, Karibib, KU289705, *, KU289720, KU289694, KU289657, discrete morphological character data. Systematic Biology 50: 913–925. KU289643, KU289682, Dactyliandra welwitschii Hook. f., DANN Maddison, W. P. and D. R. Maddison. 2009. Mesquite: a modular no. HS2434, Pandey s.n., India, Rajasthan, KU289707, KU289735, system for evolutionary analysis. Version 3.0.4. Available at http:// KU289723, KU289696, KU289659, KU289645, KU289687, Dactyliandra mesquiteproject.org. welwitschii Hook. f., DNA no. HS2435, Pandey 2201, India, Rajasthan, Meeuse, A. J. D. 1962. The Cucurbitaceae of Southern Africa. Bothalia KU289708, KU289736, KU289724, KU289660, *, KU289688, Dactyliandra 8: 1–111. welwitschii Hook. f., DNA no. HS217, Bhandari 332 (K), India, Rajasthan, Miller, M. A., W. Pfeiffer, and T. Schwartz. 2010. Creating the CIPRES Jodhpur-Balsamand, *, *, KU289722, KU289695, KU289658, KU289644, Science Gateway for inference of large phylogenetic trees. Pp. 1–8in KU289686, Dactyliandra welwitschii Hook. f., DNA no. HS1796, Carrisso & Proceedings of the Gateway Computing Environments Workshop (GCE). Sousa 3/8 (M), Angola, Moçamedes [Namibe], Rio Muçungo, KU289706, New Orleans: Gateway Computing Environments Workshop. KU289733, KU289721, KU289654, KU289646, KU289685, Dactyliandra Posada, D. 2008. jModelTest: Phylogenetic model averaging. Molecular welwitschii Hook. f., DNA no. HS2162, Menezes, Barroso & Sousa 4719 Biology and Evolution 25: 1253–1256. (LISC), Angola, Cunene, Curoza, KU289704, KU289734, *, *, KU289647, Rambaut, A. and A. J. Drummond. 2009. Tracer. MCMC Trace Analysis KU289684, Dactyliandra welwitschii Hook. f., DNA no. HS2161, Henriques Tool. v 1.5. Oxford: University of Oxford. 510 (LISC), Angola, Moçamedes, KU289709, *, *, *, *, KU289683. Sang, T., D. J. Crawford, and T. F. Stuessy. 1997. Chloroplast DNA Outgroups: Blastania cerasiformis (Stocks) A. Meeuse, M. Wilkins 279, phylogeny, reticulate evolution and biogeography of Paeonia Zimbabwe, DQ535797, DQ536656, *, DQ536803, DQ648164, *, Blastania (Paeoniaceae). American Journal of Botany 84: 1120–1136. cerasiformis (Stocks) A. Meeuse, DNA no. HS1343, Kirk s.n. (GH), Zambia, Schaefer, H. and S. S. Renner. 2010. A gift from the New World? The *, *, KU289711, *, *, KU289672, Blastania cerasiformis (Stocks) A. Meeuse, West African crop Cucumeropsis mannii and the American Posadaea DNA no. HS1340, Kotschy 205 (M), Sudan, *, *, *, *, *, KU289673, Blastania sphaerocarpa (Cucurbitaceae) are the same species. Systematic Botany cerasiformis (Stocks) A. Meeuse, DNA no. HS1341, Schweinfurth 557 (M), 35: 534–540. Yemen, *, *, *, *, *, KU289674, Ctenolepis lucorum (Keraudren) H. Schaef. & Schaefer, H. and S. S. Renner. 2011. Phylogenetic relationships in the S.S. Renner, Phillipson 2541 (P), Madagascar, DQ501260, DQ491024, *, order and a new classification of the gourd family DQ501273, *, *, Trochomeria debilis Hook f., DNA no. HS1659, Seydel 3043 (Cucurbitaceae). Taxon 60: 122–138. (GH), Namibia, *, *, KU289725, KU289697, KU289661, *, *, Trochomeria Schaefer, H., C. Heibl, and S. S. Renner. 2009. Gourds afloat: a dated macrocarpa (Sond.) Hook. f., DNA no. HS1800, Seydel 2512 (M), Namibia, phylogeny reveals an Asian origin of the gourd family *, *, *, KU289666, KU289650, *, Trochomeria macrocarpa (Sond.) Hook. f., (Cucurbitaceae) and numerous oversea dispersal events. Proceedings. Giess 13286 (M), Namibia, DQ535858, DQ536745, *, DQ536877, DQ536606, Biological Sciences 276: 843–851. *, Trochomeria macrocarpa (Sond.) Hook. f., Achigan-Dako 06nia158 (GAT), Sebastian, P., H. Schaefer, R. Lira, I. R. H. Telford, and S. S. Renner. 2012. Burkina Faso, Ouagadougou, *, *, *, *, *, AM981141, Trochomeria polymorpha Radiation following long-distance dispersal: Disentangling the con- (Welw.) Cogn., DNA no. HS2436, Schaefer HSC5 (TUM), Zambia, tributions of time, opportunity, and diaspore morphology in Sicyos KU289700, KU289729, KU289727, KU289698, KU289668, KU289648, (Cucurbitaceae). Journal of Biogeography 39: 1427–1438. KU289675, Trochomeria polymorpha (Welw.) Cogn., DNA no. HS2441, Shah, A. M., R. Tamang, P. Moorjani, D. S. Rani, P. Govindaraj, Schaefer HSC10 (TUM), Zambia, KU289701, KU289730, KU289726, G.Kulkarni,T.Bhattacharya,M.S.Mustak,L.V.K.S.Bhaskar, KU289699, KU289669, KU289649, KU289676, Trochomeria sagittata Cogn., A. G. Reddy, D. Gadhvi, P. B. Gai, G. Chaubey, N. Patterson, D. Reich, DNA no. 28755, Styles 2607 (GH), Republic South Africa, Kwazulu-Natal, C. Tyler-Smith, L. Singh, and K. Thangaraj. 2011. Indian Siddis: African inland Durban, *, *, KU289728, KU289670, *, KU289677. descendants with Indian admixture. American Journal of Human Genetics 89: 154–161. Simmons, M. P. and H. Ochoterena. 2000. Gaps as characters in sequence- Appendix 2. Representative specimens examined. based phylogenetic analyses. Systematic Biology 49: 369–381. Stamatakis, A., P. Hoover, and J. Rougemont. 2008. A rapid bootstrap algo- Dactyliandra nigrescens C. Jeffrey: KENYA. Isiolo, 4,500 ft. asl., rithm for the RAxML web servers. Systematic Biology 57: 758–771. Kirrika 34 (K). Kitui District, ca. 5 km NW of Mutomo, 850 m asl., Gillett Swofford, D. L. 2002. PAUP*. Phylogenetic analysis using parsimony. 16958 (K). Kitui, 65 km, from Mutomo on Kibwezi rd., Gatheri, Mungai, (*and other methods). Version 4.0a149. Sunderland: Sinauer Associates. Kanuri Kibui 79/161 A (K). Kitui District, 111 km from Garissa on main rd Tandon, N. and M. Sharma. 2009. Reviews on Indian Medicinal Plants to Thika, 510 m asl., Stannard & Gilbert 1108-1111 (K). Voi distr. Voi Gare, vol. 9. New Delhi: Indian Council of Medical Research. W-Pipeline M 4.5, 1,600 ft asl., Tsavo National Park, East, Greenway & Thiers, B. [continuously updated]. Index Herbariorum: A global directory Kanuri 12797 (K). Kitui District, 25 m SSW of Kitui, 3,200 ft asl., Bogdan of public herbaria and associated staff. New York Botanical Garden’s 5124 (K). Kitui District, ca. 5 km NW of Mutomo, 850 m asl., Gillett Virtual Herbarium. http://sweetgum.nybg.org/science/ih/. 16954 (K). Dactyliandra stefaninii (Chiov.) C. Jeffrey: ETHIOPIA. Sidamo, Old Airfield, ca. 15 km NNE of Yavello, 1,700 m asl., Gilbert & Jefford 4586 Appendix 1. Species and specimens sequenced for this study, (K). Sidamo, 32 km NE of Dolo, road to Bokol Mayo, 300 m asl., Gilbert, voucher specimen, geographic origin, and GenBank accession numbers Sebsebe, Vollesen 8160 (K). KENYA. Kwale distr., Kaya Muhaka (Kaya for rbcL, matK, psbA-trnH, trnL intron and trnL-F spacer, rpl20-rps12 spacer, Kambe), 30 m asl., Robertson & Luke 6294 (K). Mandera distr., Ramu- and ITS1-5.8s-ITS2 respectively. An asterisk (*) indicates sequence for Banissa Rd, 10 km from the turning to Banissa, Gilbert & Thulin 1424 (K). the subsequent region was not obtained. Herbarium acronyms follow the Turkana prov., 10 km S of road junction to Kakuma on Lokitaug-Lodwar Index Herbariorum (Thiers [continuously updated]). rd, by Lomunyenakwa river, 600 m asl., Carter & Stannard 204 (K). SOMALIA. Shabeellaha Dhexe, 179 km NE of Mogadishu on Road to Dactyliandra nigrescens C. Jeffrey, DNA no. HS218, Kirrika 34 (K), Harardere, Lavranos & Carter 23311 (K). Jalalaksi distr, 9 km S of Jalalaksi, Kenya, Isiolo, *,*, KU289713, *, KU289638, *, Dactyliandra nigrescens 125 m asl., Kuchar 17585 (K). Hiiraan reg., Bulo Burti distr., at a Bridge C. Jeffrey, DNA no. 42853, Stannard & Gilbert 1108-1111 (K), Kenya, Kitui 1.6 km N on hwy from Halgan. 190 m asl., Kuchar & Ahmed 15555 (K). District, *,*, KU289716, *, *, *, Dactyliandra stefaninii (Chiov.) C. Jeffrey, Hiiraan reg., Bulo Burti distr., 6 km N of Buli Burti Bridge on hwy. DNA no. 42854, Lavranos & Carter 23311 (K), Somalia, Shabeellaha 0.3 km W along cutline-like track. 155 m asl., Kuchar 15440 (K). Dhexe, *, *, KU289715, *, *, *, Dactyliandra stefaninii (Chiov.) C. Jeffrey, Dactyliandra welwitschii Hook. f.: ANGOLA. Moçamedes, Rio DNA no. HS219, Gilbert & Jefford 4586 (K), Ethiopia, Sidamo, *, *, Muçungo, Trepadeira, Carrisso & Sousa 3/8 (M). Loanda, Gossweiler 309 (K). KU289714, KU289690, KU289652, KU289639, KU289678, Dactyliandra Loanda, Welwitsch 833 (K). Merro da Famba?, Mendes dos Santos 1160 (K). welwitschii Hook. F, DNA no. B2516, Seydel 796 (B), Namibia, Okongawa, Southern Angola, Pearson 2641 (K). Cuanza Norte, Cacoba, Proximum KU289702, KU289731, KU289719, KU289691, KU289653, KU289640, flumen Cuanza, Gossweiler 8835 (K). 3.5 miles N of Ohopoho, De Winter & KU289679, Dactyliandra welwitschii Hook. f., DNA no. HS216, Oliver, Leistner 5646 (K). INDIA. Rajasthan, Jodhpur-Balsamand, Bhandari 332 (K). Muller, Steenkamp 6549 (K), Namibia, Weltevrede, *, *, KU289718, NAMIBIA. Omaruru, Merxmüller & Giess 1577 (M). Outjo, 37 miles from KU289692, KU289655, KU289641, KU289680, Dactyliandra welwitschii Fransfontein on Road to the Brandberg, 900–1,200 m asl., De Winter 3121 Hook. f., Giess 3664 (M), Namibia, Brandberg, 21°9′54.19″S, 14°39′39.15″E, (M). Reoboth, about 20 miles from Kalkrand on road to Rehboth (on the 72 SYSTEMATIC BOTANY [Volume 42 main Windhoek-Mariental-road), 1,200–1,500 m asl., De Winter 3537 (M). Seydel 42 (M). Shwatzor? River, Wettstein 41 (M), Namaqualand, Neue Rad Kaokoveld, 3.5 miles N. Ohopoho, slope of mountain with dolomite and und westliches Shuashwahland, ca. 1,800 m asl., Meusel 805 (M). Outjo, calcerous conglomerate, De Winter & Leistner 5246 (M). Namibrand Karbib Onava, North of Tiervlei, Crawford FC03 (K). Strohbach 2437 (K). 2115DD Okomitundu, ca. 1,400 m asl., Seydel 1275 (M). Gibeon, ridge below the Trekkopje (Grid reference), 10 km towards Okahandja from Karibib; locally Hardap dam, Giess, Volk & Bleissner 5597 (M). 8 miles West of Orowanjai abundant, creeping in grasses, Kubirske 66 (K). Karibib. Seydel 990 (K). Fountain on rd to Orupembe (Anabib), De Winter & Leistner 5645 (M). Rehoboth, Strey 2178 (K). sandy soil on banks of Kunene, long 1226, lat Kaokoveld, am Ufer des Kunene bei Otjinungua, Giess 8930 (M). 1715, Story 5795 (K). Rehoboth, Acochs 18160 (K). Kaokoveld, Ohopoho on Okahandja, Farm Erichsfelde, Padrand, Bleissner 114 (M). Omaruru, road to Anabib (Orupembe), long 1302, lat 1821, Story 5683 (K). Between Brandberg, Tsisabschlucht, Urschler s.n. (M). Omaruru, Brandenberg, Rehoboth and Uhlenhorst, Willmann 459 (K). Kaokoveld, 8 miles west of Tsisab-Valley mouth, black stony kopjes, Nordenstam 2469 (M). Omaruru, Orowanjai Fountain on the Road to Orupembe (Anabib), De Winter & Brandenberg, Tsisab-Valley, below the white Lady, in the riverbed, Leistner 5645 (K). Karibib 140 m E of Swakopmund on Kohmas highland Nordenstam 2516 (M). Farm Fahlwater: KAR 9. Unterhalb von road to Windhoek, De Winter & Hardy 7999 (K). Farm Haribes, 32 m from Granitrücken. Grober Granitgrus. 2116 CD Okabandja, Giess 11704 (M). Mariental-Maltahöhe, Tölken & Hardy 646 (K). 2116 CD Okahandja (Grid Berghang, Fraventzheim, Leippert 4630 (M). Outjo, Farm Winkelhaag (OU ref.), Farm Fahlwater, KAR 9, Giess 11704 (K). Swakopmund, crossing of 286); dolomit mountain, lower slope, Merxmüller & Giess 30364 (M). Omaruru river, N of Hentjes Bay, Wanntrop 246 (K). 1712 BA (Grid ref.), Brandberg, Southeast side c. 5 miles South Tsisab gorge, Giess 3664 (M). Kaokoveld, Otjomborombonga on Kunene River, 360 m asl., Leistner, Oliver, Kaokoveld, gully 77 miles from Ohopoho on road to Anabib (Orupembe), Steenkamp, Vorster 114 (K). 2415B (Grid ref), Weltevrede, NE of Sossusvlei, long 1302, lat 1821, Story 5683 (M). Sandy soils on bank of Kunene, long Oliver, Muller, Steenkamp 6549 (K). 1613 CD (grid ref.), Kaokoveld, 6 km e 1226, lat 1715, Story 5795 (M). Gibeon, Launert 2285 (M). Gobabis, Farm of epupa falls, 620 m asl., Leistner, Oliver, Steenkamp, Vorster 269 (K). 2215 Donnersberg, Volk 1685 (M). Oterhandja, Erichsfelde, Volk 1889 (M). CD, Bloedkoppie, just w of Tinkas in Namib desert park, Oliver, Muller, Windhoek, Farm Otjisewa, Kinges 2484 (M). Farm Nudis bei Karibib, Steenkamp 6592 (K). Okahnja, 1,300 m asl., Dinter (K).

View publication stats